No human studies have systematically examined the relations among PaCO2, cerebral blood flow, and the cerebral metabolic rate for oxygen during hypothermic cardiopulmonary bypass. We varied PaCO2 during hypothermic (26-28 degrees C) cardiopulmonary bypass and estimated the cerebral metabolic rate for oxygen by multiplying cerebral blood flow (measured using xenon-133 clearance) by the cerebral arteriovenous difference in oxygen contents. Patients were randomly assigned to either of two methods of managing PaCO2 (uncorrected for body temperature). In group 1 (PACO2 32-48 mm Hg, n = 13) the mean +/- SD cerebral metabolic rate for oxygen was 0.40 +/- 0.11 ml O2 X 100 g-1 X min-1 at a mean +/- SD PaCO2 of 36 +/- 2.0 mm Hg and 0.40 +/- 0.14 ml O2 X 100 g-1 X min-1 at a mean +/- SD PaCO2 of 45 +/- 2 mm Hg. and 49-72 mm Hg, n = 12) the mean +/- SD cerebral metabolic rate for oxygen was 0.31 +/- 0.09 ml O2 X 100 g-1 X min-1 at a mean +/- SD PaCO2 of 55 +/- 3 mm Hg and 0.21 +/- 0.07 ml O2 X 100 g-1 X min-1 at a mean +/- SD PaCO2 of 68 +/- 2 mm Hg. Group 2 values differed significantly from those in Group 1 (p less than 0.05). In both groups, cerebral blood flow increased as PaCO2 increased. During cardiopulmonary bypass, increasing PaCO2 increases cerebral blood flow and decreases the cerebral metabolic rate for oxygen.